Network Storage Protocols Articles and Resources

Performance Characteristics of Datastores Accessed via FCP and NFS on VI3.5


This paper documents the performance characteristics of Virtual Infrastructure 3.5 datastores accessed

via FCP and NFS protocols. Testing methodology, configuration, and results will be presented for

connectivity via 4Gbps FC, 1GbE, and 10GbE. The results in this paper show that the FCP and NFS

protocols are both viable options for high performance connectivity to shared storage in a Virtual

Infrastructure 3.5 environment.

Please Note:

All content posted on the NetApp Community is publicly searchable and viewable. Participation in the NetApp Community is voluntary.

In accordance with our Code of Conduct and Community Terms of Use, DO NOT post or attach the following:

  • Software files (compressed or uncompressed)
  • Files that require an End User License Agreement (EULA)
  • Confidential information
  • Personal data you do not want publicly available
  • Another’s personally identifiable information (PII)
  • Copyrighted materials without the permission of the copyright owner

Continued non-compliance may result in NetApp Community account restrictions or termination.


Realy cool, but you place only 2 VM's in one VMFS. It becomes more interesting when you place 10 or 15 VM's in one VMFS (the real world and like you do in the nfs store).

You will see then some difference, in the advantage of nfs.





That's a great observation, I didn't even notice that the FCP datastore only had 2 VMs per.

I'm going to check with Mike Arndt (the engineer who ran the benchmark) to see why this was done? and if he had any test results with running "10 VMs on NFS" vs "10 VMs on FCP"?

Are you running NFS in your environment? Have you done any benchmarks?

Cheers, Tony



Are the bandwidth / overhead of the three protocols reasonably comparable?  For example, I have ESX hosts that are using dual-pathed 2Gb FCP ports, but according to the FC switch, no single port reaches more than 40 Mbps peak utilization.

If the bandwidth numbers are comparable, assuming we put the proper infrastructure in place with regards to reliability, from a throughput perspective, we could in theory consolidate these ESX servers to just a handful of 1 GigE ports without incurring performance penalty?



Sorry, but I just noticed this document out here and the questions posted.  There are not only 2 VM's in each VMFS datastore.  There are 2 VMs *per server*, on each of the 5 servers, per VMFS datastore.  This means there are 10 VMs per VMFS datastore in total, as each datastore is connected to all 5 servers in the ESX cluster.  Hopefully this helps clear things up if people are still confused on this point!

All Community Forums